Chapter IV, Part 1
IV.A.
Form, meaning, function, and textuality
1.
It might sound like a paradox to suggest that a functional
description of a language is more likely to attain °convergence and
consensus° than is a formal description.
After all, aren’t the forms ‘right there on the page’ when you write them
down? Wouldn’t people agree much better about what forms they see than about
what functions are being served? But this disparity in consensus is misleading,
due largely to the visual enhancement of standardized orthography and the
‘cheap explanatory work’ it can do (cf. II.24, 29, 34, 38; III.195, 202).
Actually, the linear sequence by itself is too sparse. The skilled procedures of
literate native speakers are merely more mechanical
and simple for writing down words and
phrases and yield more tangible results than do their skills for assigning them
functions, which are more creative and
complex.
2.
For similar reasons, °formalism° in the study of language or discourse looks
like a more straightforward and rigorous way to analyze data than °functionalism°.
But the advantage holds only if we work within small, closed circles. Our easy
consensus about forms demands that we stick
to the surface forms — by segmenting them and classifying the pieces (as
in ‘immediate constituent analysis’), or by shifting them around (as in
‘transformational grammar’), and so on (cf. II.34, 42). Consensus drops
sharply when we try to describe where
the forms and their patterns come from
and why speakers say them one way rather than some other way. By
rejecting functional explanations, formalism implies the decidedly unrealistic
notion that speakers use language in order to arrange forms in well-formed
strings but can’t quite manage it (III.172).
3.
So the aspirations of formalism to be the mainstream and majority position in
linguistics have been supported by an unfair and simplistic advantage that
eventually defeats rather than promotes °coverage, convergence, and consensus°
(II.57ff). Work has to concentrate on the more °frozen aspects or sectors of
language°, e.g., by distinguishing whether a language predominantly has a
‘Subject-Verb-Object’ pattern or a ‘Object-Verb-Subject’ pattern but not
by defining a cognitive or social conception of ‘Subject’. Also, we are
encouraged to display the tips of these ‘frozen icebergs’ and to keep the
functional aspects of our analysis out of sight, which are therefore hard to
compare and negotiate. Naturally, such outlooks maneuver us into the position of
trying to °freeze things into ‘structures’ and ‘rules’ that aren’t
frozen in authentic data°; and the results tend to be too °arbitrary and
unsystematic to support consensus° (II.47).
4.
Conversely, functionalism was for a time fended off from the mainstream into a
minority position by an unfair disadvantage: the convergence and consensus about
functions are easily attained for discourse communication but are not displayed
as simply and directly as those about forms (II.61). Moreover, language is
‘functional’ in several distinct senses that richly interact and resist tidy
segmentation and classification. Whereas forms can be described by relations of part
to whole or left to right,
functions need to be described by relations of means
to end (II.59ff). And whereas the parts and wholes to the left or the right
are ‘there on the page’, the means and ends are only implicitly and
partially represented, and require °rich mediation though the intentions and
actions of speakers and hearers and through their knowledge of world and society°.
So the procedures of functionalism are seldom cut and dried, nor are its results
likely to attain some reassuring completeness and finality. We are continually
in the midst of ‘work in progress’, attaining a richer understanding of
language in action but never finalizing or exhausting it.
5.
The richness extends far beyond the °levels
of language that linguistics conventionally defines by their theoretical formal
units° (cf. II.30, 59). Following the Prague school, I have proposed a functional account wherein the forms are the means for the ends of
meaning; also, the sparser forms, e.g. morphemes as word-parts, are the means
for the ends of the richer forms, e.g. lexemes as words and syntagmemes as
phrases (II.60ff). From there we can develop a °systemic functional account wherein the formal resources are multi-functional°
and comprise a systemic network of means that can serve a vast range of
ends, some of them not yet realized. °Authentic
data° in context may confront us with fluctuating, novel, and complex
arrays of means and ends, and with a richness that far outstrips our customary
terms and methods of analysis and description. In contrast, °invented data° whose sole function is to illustrate presumed
‘grammatical rules’ imply a fictional mono-functionality
that seems precise (determinate) only if we stay on the surface; when we go
deeper, the data get fuzzy (indeterminate) because crucial constraints have been
dispersed. This factor explains why invented data are typically trivial and
simple-minded, e.g., ‘the man hit the ball’ or ‘John is eager to please’
(II.47, 66; IV.36, 57, 82): this tactic conceals how much is left out, and fits
the trivial picture of language that formalism is forced to provide.
6.
When the relation among ‘levels’ is interpreted in terms of means and ends,
then their respective theoretical units — phoneme, morpheme, lexeme, and
syntagmeme — cannot be static forms,
but rather dynamic form-function
connections. For the forms we actually see as stretches
of text, we can use the terms for ranks
like those envisioned in British functional linguistics: Sound/Letter - Syllable
- Word - Collocation - Phrase - Clause - Clause Complex - Sentence/ Utterance -
Discoursal Move - Discourse Episode - Text - Discourse. Here, the practical
part-whole relations are largely stipulated by the ordinary senses of the terms
for ranks — e.g., a Syllable being a part of a Word and being pronounced as a
unit, or a Phrase being a sequence of Words — and hold no profound theoretical
significance. Our theories and methods need not contend with the problems that
troubled formalist linguistics, such as ‘discontinuous constituents’ or
fuzzy borders between the Word and the Phrase (cf. II.59). Nor do serious
problems arise when a unit occupies more than one rank, e.g., a Word being also
an entire Utterance; or when a unit gets ‘rank-shifted’, e.g., when the
Direct Object in a Clause is itself a Clause (cf. II.59; IV.227). The ranks
merely provide °practical terms with heuristic functions°
for promoting a °convergence of data and a consensus among investigators° more
readily than would °theoretical terms
with strictly formal definitions° (cf. IV.20ff, 30).
7.
Now that functionalism is arriving in the mainstream, it’s time to assess how
and how far we can still apply formalist approaches. Trying to build a
functional description directly on top of a formal description is problematic.
The two modes of description may well not arrive at matching categories or
patterns, because the relations between form and function are seldom frozen or
one-to-one. Some of relations are fairly well-defined by °standing constraints°,
e.g., the function of the morpheme for the English ‘Plural’ usually written
‘‑s’ or ‘‑es’, or the function of the lexeme for
‘aardvark’ in the discourse of zoology. But most relations fluctuate to suit
°emergent constraints° as well, e.g., the multiple functions of ‘on’ that
permit the amusing but obviously unintended readings of headlines like [59-61].
[59]
City, county, Union Pacific meet on radioactive soils (Las Vegas Sun,
6/7/ 1985)
[60]
Jury is still out on composting toilets (Salem Statesman-Journal,
15/3/1981)
[61]
Internal memos on tampon introduced (Washington
Post, 8/4/1982)
Yet
a multi-functional item like ‘on’ is not harder
to use than a fairly mono-functional one like ‘aardvark’, although
ordinary speakers of English could hardly give an explicit description of the
functions of ‘on’ (cf. I.59; III.40; IV.152).
Evidently, discourse participants can fit forms to functions, and can fit the
means to the ends, not by combatting
fluctuation, novelty, and complexity but by exploiting
them in a supportive rather than resistive
mode (cf. I. 40; III.42ff; V.55). This capacity should be better mastered by us
linguists, who are still combatting them because we’d like our description of
language to be stable, familiar, and simple.
8.
In Ch. III, language and its meaning were described as eminently °non-classical
phenomena, merging concurrent
alternatives and local interactions
to reach the transitory but potent critical mass of context°. Discourse does
not seek total certainty or determinacy,
but rather a °convergence of constraints
making some meanings or understandings reasonably
probable° (cf. III.142, 248). The same holds here for the connections
between form and function. In discourse, a succession or pattern of language
forms is assigned probable functions, using available cues from knowledge of
world and society and from the ongoing situation wherein the total discourse is
‘functioning’. Some functions may be rather unusual and statistically
improbable in respect to the whole language. In lines 17-19 of Coleridge’s
celebrated poem ‘Kubla Khan’:
[62]
And from this chasm, with ceaseless turmoil seething,
As
if the earth in thick fast pants were breathing,
A
mighty fountain momently was forced
the
Noun ‘pants’ as a Plural Form for the Action of ‘panting’ (nearly always
expressed by a Verb) is quite rare and marked in general English, while the Noun
‘pants’ is quite common and unmarked for underpants in British usage and for
trousers in American usage; but the context reverses these probabilities. Yet my
American students in a pilot project for an ‘Introduction to Poetry’ class
(VII.300ff) have sometimes read ‘pants’ for trousers and even exploited this
for an erotic interpretation of items like ‘chasm’ — evidently a possible
assignment of functions which the routines of conventional literature classes
renders rather improbable.
9.
Moving from formal toward functional also dominates the agenda of text
linguistics (cf. II.F). Projects for describing texts as formal units made of
rows of sentences soon foundered because the text is primarily a functional
unit. So we shifted our focus from °text grammar over to textuality°, as
described in II.105-11. But we did
not yet go far enough. The seven principles of textuality surveyed in the 1981 Introduction
to Text Linguistics and informally presented in section I.C of this book
still need to be reinterpreted from a °transdisciplinary
standpoint°. Otherwise, we still rely too much on the
‘divide-and-conquer’ methods of linguistics for °segmenting and classifying
the forms of texts from the bottom up°. Doing so encourages us to treat each
principle of textuality in isolation from the others and to match up the
principles with the mainstream linguistic schemes
of ‘levels’ or ‘components’,
e.g.:
cohesion = morphology, syntax, grammar
coherence
= semantics
intentionality
+ acceptability + situationality = pragmatics
informativity
= topic/comment, theme/rheme, intonation
intertextuality
= stylistics
Such
a matching is inappropriate insofar as the older levels were described mainly in
°formal terms and in mutual isolation for language by itself as a virtual
system°, whereas the newer principles should be described mainly in °functional
terms for the text as an actual system°.
10.
Today, our agenda calls
for interfacing functionality with textuality, while formulating
projects that can serve the ecological
goals of our science (cf. III.186; IV.41, 138). A formal description of
language by itself has usually been construed as a theoretical goal in its own
right, probably because further goals were hard to imagine. In contrast, a
functional description is more properly a means
to some practical goal, typically to explore socially significant strategies
and problems in human discourse (I.33, 59f; III.186). It would be inappropriate
to judge the two modes of description by the same standards, especially if those
standards are to be set by the formalist side, e.g., ‘the more formal, the
better’ (cf. II.57), where we might more aptly say: ‘as formal as necessary
but as functional as possible’. (as in II.79) Functional descriptions are to
be judged by how far they enhance °coverage, convergence, and consensus about
corpuses of authentic data and about the cognitive and social issues relevant
for discourse practices° (II.75, 132).
11.
A °lexicogrammar° along these
lines would have three correlated aspects. It would be expressly (a) functional
in exploring how language resources are typically used, e.g., Modal Verbs for
polite requests; (b) cognitive in
indicating how those resources are constrained by the community’s shared
knowledge about the organization of Processes and their Participants in world
and society, e.g., what Actions are sensible to request; and (c) social in highlighting the constraints upon discourse interaction,
e.g., who is legitimized or empowered to make requests and commands (Section
IV.B). These three aspects, roughly corresponding to Halliday’s three
‘metafunctions’ of ‘textual, ideational, and interpersonal’,
respectively, are needed because — to the dismay of some linguists and
logicians — a lexicogrammar is a social construct and reflects a
commonsensical cognitive view of world and society rather than a mathematical,
logical, or scientific analysis. Instead of a strict boundary between what is
versus is not ‘grammatical’, we find relative degrees of markedness,
ranging from unmarked for routinely
chosen general options over to highly
marked for options with special motives and effects (cf. I.19; III.235;
IV.8). Degrees of markedness also apply to all three aspects, as we can see in
this passage from The Book of Mirdad,
by the distinguished Lebanese writer Mikhail Naimy. The narrator has taken
refuge at night in a grotto after climbing a rocky mountain slope, only to be
wakened by ‘a very old man’ who takes away his staff and gives it to ‘a
woman as old as himself’:
[63]
Then as if taking note of me, but always speaking to his companion, he added
‘The stranger shall depart anon, and we shall dream our night’s dream
alone.’ This fell upon me as a command which I felt impotent to disobey,
especially when the dog approached me snarling menacingly as if to carry out his
master’s order. […] ‘My staff you have taken. Will you be so cruel as to
take this grotto also which is my home for the night?’. ‘Happy are the
staffless, they stumble not; happy are the homeless, they are at home.’
We
notice some marked grammatical options of language in ‘fronting’ parts of
the Predicate before the Subject, e.g., the Direct Object Target (‘My staff
you have taken’) or the Emotive State (‘Happy are the staffless’). But we
can also notice the cognitive markedness of being, say, both ‘homeless’ and
‘at home’; and the social markedness in the aggressive behavior of the old
man in depriving a stranger and turning him out onto an unknown and dangerous
terrain in utter darkness. We can then appreciate the total marked effect, which
the narrator models for us: ‘the whole scene filled me with terror; I watched
it as in a trance’, ‘making desperate efforts to speak — to defend myself,
to assert my right’. Significantly, ‘speaking’ is invoked as a means for
managing the situation in a more humane direction but does not work because the
old man will not accept an unmarked role in a conversation or even address the
narrator directly (and the implied ‘command’ is enforced by a ‘snarling
dog’).
12.
In this chapter, the
three °functional levels° proposed in II.63 will be elaborated in several
stages from a largely systemic viewpoint. For the level of °lexicogrammar°,
we can consider the typical language resources whereby major Aspects
and Processes are lexicalized and grammaticalized. Then, we can inquire
how lexicogrammatical resources can be maintained
and compacted in discourse by
reusing or compressing the patterns that are already produced (IV.C); and how
patterns can be joined (IV.D). For the
level of °prosody°, we can explore
the intonation of spoken discourse and
the punctuation of written discourse (IV.E).
Finally, we shall regain the level of °discourse°,
where all the resources and constraints converge in a context of situation (IV.F).
IV.B.
Toward a functional, cognitive, and social lexicogrammar
13.
The term lexicogrammar
was made current by °systemic functional linguists° like Michael Halliday, who
has called it ‘the inner core of language’, to designate a unifying
perspective on a richly interconnected range of language phenomena. In its own
way, each language exploits the two sides to share the ‘work’ of expression,
sometimes °grammaticalizing° more,
as in ‘polysynthetic’ languages like Aymara (II.31), and other times °lexicalizing°
more, as in ‘analytic’ languages like English (II.62). The functional unity
of the two sides can be seen when lexical items entail grammatical constraints,
while grammatical patterns prefer certain types of lexical items — a vision
staunchly reinforced by large-corpus linguistics (II.75, 78). The unity is
indicated by empirical findings as well. Speakers can either select the lexical
items first and then arrange them in grammatical phrasings, or else can select
phrasings first and then the lexical items to fill them (cf. III.237). And when
infants shift from spontaneously designed sounds over to real lexical items, the
grammar soon begins assimilating to the mother language (III.130; VII.145).
Still, the two sides clearly differ in their evolution, with the lexicon
changing faster, accepting more deliberate innovations, and forming less
systematic classes than does the grammar (III.118).
Also, the lexicon has a far more diverse and multiple range of functional
orders, reflecting the normally improvised ways in which cultures use vocabulary
to express and classify objects, events, and actions (cf. IV.2).
14.
When linguists separate
‘lexicon’ from ‘grammar’ and concentrate on the ‘grammar’ side, as
most formalists have (II.33, 49), they are being attracted toward the side where
stability and simplicity appear much higher. Two contrary tendencies can result:
either to put too much into the
grammar as ‘rules’ drawn from examples that are in fact partly constrained
by their lexical items; or else put too
little into the grammar by excluding all cases where lexical constraints are
suspected to apply. Traditional grammars have favored the first tendency,
whereas formalist grammars have favored the second. Our safest recourse is to
suspend the project of separating grammar from lexicon and to use a large corpus
of authentic data to determine how the two sides share the ‘expressive work’
in a given language or discourse domain (cf. II.75ff).
15.
Ironically, English —
the very language for which the most formal grammars have been devised —
lexicalizes far more than many languages, and this trend increases as the
language evolves. So a unified functional lexicogrammar should be particularly
useful for English. Obviously, it cannot be a traditional grammar issuing °prescriptions
for ‘correct’ and proscriptions for ‘incorrect’ usage° and thereby
joining a crusade to deny the realities of usage (VII.226ff). Nor can it be a
formalist grammar for normalizing data and converting them into arrays of
formulas without asking why people say what they do (II.41). Instead, we can
work out a scheme of Processes richly
connected to cognitive and social knowledge about how events happen or who does
which actions. Each Process can be handily represented by at least one °lexicogrammatical
Prototype° within a cluster of less common or more specific lexical items,
e.g., ‘doing to’ for the Dispositive Process (realized also by
‘seizing’, ‘eating’, ‘destroying’ etc.) versus ‘knowing’ for the
Cognitive Process (realized also by ‘believing’, ‘guessing’,
‘imagining’ etc.). Each Process is formally distinctive in at least some of
its Aspects but by no means in all. And the main distinctions lie in the
distribution not of grammatical versus
ungrammatical sentences but of unmarked
versus marked collocations of
co-texts constrained by both grammar and lexicon. In English, the range of
marked options is much sparser for the grammar, e.g., °fronting the Object°
(as in ‘My staff you have taken’ from [63] in IV.11)
and much richer for the lexicon, e.g., making new compounds (as in
‘staffless’ from [63]), so that a rather small set of grammatical patterns
for phrases and clauses does a big portion of the work. This proportion is
skewed still further when some Prototypes like ‘doing to’ or giving to’
function as °attractors° for other
Processes we must specify by lexical means, e.g., ‘strike an agreement’
patterned after the Dispositive of ‘doing to’ but specified by its lexical
items to be a Semiotic Process (cf. IV.64).
16.
This high sensitivity of
English grammar to lexical control is a natural but disheartening barrier
against a stringent formal description. As we have seen for large-corpus data
(II.68-79), lexical control upon grammar is an indispensable resource for °coverage,
convergence, and consensus°. But acknowledging this control obliges us to
exercise great care and discretion while applying our categories to °authentic
data°, where the formal criteria and distinctions often prove insufficient by
themselves. When our lexicogrammar is gradually placed upon the foundation of a
large corpus, as Michael Halliday has
foreseen (IV.165), the design will doubtless be modified under pressure from the
data, particularly by acquiring much greater °delicacy
in the sense of Halliday and Hasan, revealing rich grammatical control on the
lexicon and rich lexical control on the grammar° (cf. II.65, 77f; III.41;
IV.29).
17.
These issues lead to the
long-standing question of what format
should be used to represent language data.
In formalism, the consensus has chiefly been that °language data should be
rewritten into a formal notation for purposes of analysis, description, and
explanation°; yet as the data get made sparser, the results of various
‘formalizings’ become less likely to converge (III.167-70). In
functionalism, the consensus is rather that we should stay close to the actual
data and let them represent themselves (IV.19).
Notations should be only an auxiliary
medium for other tasks, such as the ‘proposition lists’ used by
cognitive psychology to assess units of processing and recall, or the bracketed
‘LISP’ programs used in artificial intelligence for computer simulation of
certain operations in natural language. Both strategies support convergence and
consensus: the data by retaining rich constraints, and the notation by
enabling ‘higher control’ across whole sets of data, e.g., in order to make
comparisons and measurements when assessing which versions of a text are better
recalled (V.26ff).
18.
In the 1981 Introduction
to Text Linguistics, Dressler and I proposed a compromise notation that
stayed close to data while making the types of relations more explicit. So we
represented °cohesion
and coherence° in
two parallel
networks, one with ‘grammatical’
links among words
at
the nodes
(e.g.
‘head’ and ‘modifier’) and one with ‘semantic’
links among concepts at the nodes
(e.g. ‘Attribute’ and ‘Location’), as suggested in Fig. IV.1
for the sentence ‘a black and yellow rocket stood in a New Mexico desert’.
We finessed the problems of relating grammar to lexicon by retaining the words
as labels for the concepts, on the grounds
that no other
system of labels devised so
far has attained enough range and consensus to replace the words, which are at
least privileged by their occurrence in the ‘surface text’. Networks have
the representational advantage of
making linkages visually richer and more explicit than do the linear order of
words and the sparse linkages of constituency. But their operational
properties may be even more significant if, as suggested by research cited in
Chs. II and III, the on-line organization of discourse processing might be
managed through °self-organizing networks arising from the activation and
integration of local data°. If such operations exploit the prior organization
of knowledge about language, world, and society all at once, then we would have
not just a series of deterministic and stable networks close to the ‘surface
text’ like those shown in Fig. IV.1,
but a single evolving ‘multi-network’
connected to knowledge that was partly ‘expressed’ by the text and partly
supplied by ongoing processes such as spreading activation, inferencing, and
updating (III.246). The nodes shown in the Figure would only be cues or
instructions within a much richer and more transitory array. How such an array
might be represented in a consensual graphic pattern is of course far from
settled. Though most functionalists would grant that the ‘surface text’ as
we see it written down on a page is rather like the mere tip of an immense
iceberg seen sticking out of the water (I.34, 58; IV.3), we remain cautious
about declaring what the ‘unseen’ portion might look like. And if our
networks are obvious idealizations, much work must be done to design more
‘realistic’ ones.
19.
Meanwhile, we can make
some headway in describing the lexicogrammar by working with the ‘surface’
patterns of phrases and clauses in real discourse and sorting them according to
how they °grammaticalize and lexicalize patterns of knowledge about world and
society, thereby interfacing linguistic constraints with cognitive and social
ones°. For most purposes, we can let the data represent themselves rather than
transposing them into some other notation or network whose sparseness might
obscure their ‘naturalness’, much as a °critical dispersion loses key
emergent properties° (cf. III.169). This way, the lexicogrammar may be °user-friendly°
and also adaptable to specific °ecological goals°, such as exploring the
discourse of education (VII.I).
20.
We can also retain our
familiar schemes for °Parts of Speech°,
which have always mixed formal with functional categories, especially when a
grammar for a formally rich language like Latin got applied to a much sparser
language like Modern English. A Latin form like ‘lumen’ is plainly a
‘Noun’ with a distinct formal repertory (or declension) of ‘cases’ (e.g.
Genitive ‘luminis’, Dative ‘lumine’); an English form like ‘light’
might be a Noun (in all its ‘cases’), a Verb, an Adjective, or an Adverb,
depending on its function in a Clause (cf. IV.28). Mixing form with function and
extrapolating from the grammar of one language to another have drawn widespread
criticism against Parts of Speech schemes, but attempts to discard or replace
them altogether with fully formal schemes have not supported °coverage,
convergence, and consensus° (IV.3). So, like the °ranks such as Syllable or
Word°, the Parts of Speech can be retained as practical terms with heuristic
functions rather than as theoretical terms with formal definitions (cf. IV.6,
30). They too can be capitalized for easy recognition.
21.
In this heuristic
spirit, the ratio between theoretical versus practical can be aligned with the ratio
between virtual units versus actual units. The Sound
would be any actual sound of a
language, e.g. a Vowel, Consonant, or Diphthong, that corresponds to a virtual phoneme and can
distinguish between two Words which also differ in meaning. The Word-Part
would be any actual part of a word,
e.g. a Stem, Prefix, Suffix, or Infix, that corresponds to a virtual morpheme and can
appear as a recognizable component occupying a regular position within each of a
group of Words and contributing to their meaning. The Word would be any actual
lexical item, e.g. a Noun, Verb, or Modifier, that corresponds to a virtual
lexeme and can appear as a self-contained unit with a relatively
stable internal ‘form-base’ and a relatively fluctuating external position
inside the Phrase. The Phrase in turn
would be an actual word-pattern, e.g.
a Noun Phrase, Verb Phrase, or Prepositional Phrase, that corresponds to a virtual
syntagmeme and has positions for specified ‘Word-Classes’ or ‘Parts of
Speech’. In languages with alphabetic writing systems, the Word is generally
written as one unit and remains unvaried or undergoes a sparse, deterministic
range of regular changes of its Word-Parts, while the Phrase is written as more
than one unit, varies more freely, and undergoes a richer, less deterministic
range of insertions or expansions; but for functional motives, many items or
combinations can hover along the border between Word and Phrase. Aligning the ratio between theoretical versus
practical with the ratio between virtual versus actual for the units
on levels as well as for the whole
system of language and discourse enables a linguistics of text and
discourse to readily integrate the results of descriptive linguistics, where the
ratio between theoretical versus practical was sometimes problematic (II.24).
But we must not forget: it is the linguists
who are identifying and locating theoretical units like phonemes, morphemes,
lexemes, and syntagmemes, where real speakers
and hearers would attend to the practical units like Sounds, Word-Parts,
Words, and Phrases. And how far the linguists are acting in ways similar to real
speakers and hearers is a significant question for both theory and practice and
should not be short-circuited by a glib or hopeful terminology (cf. II.26, 36,
46, 75).
22.
The Word-Class
or Part of Speech would be a set of
Words or Word-like units that assume similar forms or positions (or both) and
that fulfill prototypical parameters of functions (cf.
II.35, 69). A useful heuristic distinction has been made in English between
classes of Function Words (usually
Articles, Prepositions, and Conjunctions) versus classes of Content Words (usually Nouns, Verbs, and Modifiers) (cf. II.35, 69;
III.208). These two terms designate the poles of a gradation between small
Word-Classes having sparser, fluctuating meanings and adapting strongly to their
environments (e.g. ‘on’), versus Word-Classes having richer, more stable
meanings and adapting weakly to their environments (e.g. ‘aardvark’) (cf.
IV.7). Evidence for this gradation, cited back in II.35, is that English
speakers are far less likely to utter a Function Word alone (e.g. ‘the!’,
‘at!’, ‘until!’) than a Content Word (e.g. ‘fire!’, ‘run!’,
‘terrible!’). Yet the two terms are slippery, since all
Words have at least some ‘function’ and ‘content’; the distinction
designates relative priorities in the
one direction or the other. Further specifications readily occur within the °economy
of a discourse°, e.g., Wallace Stevens’ ‘intricate evasions of as’ (from An
Ordinary Evening in New Haven), where a Function Word ‘as’ appears in a
Prepositional Phrase like a Content Word (e.g. like ‘simile’, cf. VII.296.)
.
23.
For a functional
description of English, a system of Word-Classes or Parts of Speech would be °non-deterministic°
in at least seven aspects. First, °lexical
control° keeps most regularities from being purely grammatical in the sense that they could apply to all
instances of the same formal categories all across the language and to no
non-instances. We will seldom find patterns of a ‘Noun Phrase’ or a ‘Verb
Phrase’ in which any Noun or any
Verb could equally well be inserted. Instead, we often find certain combinations
of ‘Noun Phrase’ + ‘Verb Phrase’ which resist
certain lexical classes of ‘Nouns’ or ‘Verbs’. For example, not all
Verb classes can be used in Passives, such as ‘self-directed corporeal
Enactments’, e.g., ‘he blew his nose’ =>
‘??his nose was blown by
him’ (IV.89). In return, lexical control can also improve upon otherwise
doubtful patterns, e.g., ‘after all the children’s noses had been properly
blown, they went down to meet their new teachers’ (implying an adult had
primly circulated with hankies) (cf. IV.89). This capacity of the lexicon to
substantially modify our judgements of grammatical patterns, which we will be
seeing again (IV.50, 70f, 75, 107), has been another major obstacle to formalist
descriptions that want everything decided in the grammar alone. Evidently, a
general grammatical constraint neither is a natural law nor is it refuted merely
when we can deploy the lexicon to devise counter-examples.
24.
Second, the meaning
of a Word-Class is only °prototypical°.
We can always expect some exceptions to a definition, whether it is more
commonsensical, e.g., a noun being
‘a person, place, or thing’, or more technical, e.g., a noun
being ‘the name of a subject of discourse (as a person, animal, plant, place,
thing, substance quality, idea, action, or state)’ (Webster’s
Dictionary, p. 577), where the definition begins to encroach on the
definition of the verb
(‘action’) or the Adjective (‘quality’). Also, creative discourse finds
new ways to ‘Nominalize’ an Action (e.g., when the Action of seeing is
termed ‘the sensible and true avouch of mine own eyes’, Hamlet
I, i, 57-58) or to ‘Verbalize’ an Agent (e.g., ‘England’ ‘is so idly
king’d’, Henry V, II, iv, 24-26).
Confusion seldom arises unless we mix up Word-Class with meaning and make
statements like ‘a noun performs an
action’ ([755] in VII.260) or ‘the Verb acts upon its Object’, as if
grammatical categories could do things out in the real world.
25.
Third, a single item can belong to
more than one Word-class, depending on the context (cf. IV.20). Though many
English Words, like ‘visit’, ‘waffle’, ‘plan’, etc. can be either noun
or verb, contexts usually specify which grammatical pattern is
probably intended and which is not, as in:
[64]
Reagan Visits Harassed Blacks (San
Francisco Chronicle, 4 /5/1982)
[65]
British left waffles on Falklands (The
Guardian, 28/4/1982)
[66]
Large church plans collapse (Hamilton [Ontario] Spectator, 8/6/1985)
To
argue that such instances are mere coincidences wherein ‘different’ Words
appear the ‘same’ in sound (‘homophones’) or in writing
(‘homographs’) is to discount associations of meaning (e.g., for both uses
of ‘visit’) and parallels of form (e.g., the conjunction
‘until’ in ‘until she arrives’ and the preposition
in ‘until her arrival’). And we get a disjointed picture of how verbs
get ‘Nominalized’ and nouns
get ‘Verbalized’.
26.
Fourth, some sets of
items hover between Function Words
and Content Words (in the sense of IV.22). In between Pro-Nouns and Nouns,
some items are Nouns by form but have sparse and fluctuating meanings, such as
‘thing’; in spoken Italian, ‘cosa’ for ‘thing’ has indeed been
shortened from ‘che cosa?’ to become a Function Word Interrogative for
‘what?’. In between Auxiliaries and
Verbs, the ‘Pro-Verbs’ like
‘do’ and very general Verbs like ‘act’ also fluctuate; and some
non-Modals closely resemble Modals in meaning, e.g., ‘is capable of’ vs.
‘can’, or ‘is supposed to’ vs. ‘should’. In such instances, our
description should be flexible in recognizing unified functions despite
disparate forms.
27.
Fifth, a standing item
in a Word-Class can be composed of more
than one Word (e.g., the English verb
‘look at’ vs. Spanish ‘mirar’, or the preposition
‘instead of’ vs. German ‘statt’). Here we find the °frozen combinations°
usually called Idioms (like ‘make
an ass of yourself’) and the more flexible but still predictable combinations
that J.R. Firth called Collocations
(e.g. ‘silly ass’, his favorite) (cf. II.67). The drift toward
‘freezing’ is signalled in two ways. The meaning of the whole becomes quite
distinct from what its parts would literally suggest by themselves (e.g.,
transforming oneself into a donkey). And the forms become stabilized and
restricted in range, as when a Verb phrase
loses some of its repertory (e.g., ‘spoiling for a fight’ being constrained
to Progressive
Tenses,
whence the oddness of Simple Tenses like ‘?he
spoiled for a fight last week’). Or, a noun
phrase can have a modifier
whose pattern resists internal inflections or insertions (e.g., ‘new age’ in
‘new age music’ could hardly be changed to ‘?the
newest age music’ versus ‘the latest new age music’, or to ‘*new bland
age music’ versus ‘bland new age music’).
28.
Sixth, the Word-Classes need not specify every item in a clause or sentence or assign
it an exact formal relation to the other items. We can expect some instances
where more than one description is possible, e.g., whether a Participle like
‘participating’ in ‘participating outlets’ is a Verb or Modifier; or
whether ‘bright’ in ‘bright before mine eyes twin trumpeters stand’
(Edward Thomas) is an Adjective for ‘trumpeters’ or an Adverb for
‘stand’; or whether ‘in’ in ‘I found a bed to sleep in’ is an Adverb
or a Preposition. Similarly, the division into ‘Subject’ and ‘Predicate’
may leave a residue of Adverbial or Junctive items (e.g. ‘frankly’, ‘in
consequence’) whose function relates to the ongoing discourse rather than to
some Phrase Head within the Clause (cf. II.38; IV.206).
29.
Seventh, a functional
lexicogrammar remains incomplete.
Halliday has remarked that ‘anything approaching a complete grammar would be
hundreds of times the length’ of his own 434-page Introduction.
Obviously, such a project couldn’t be compiled by a single investigator nor
published in the usual book format; we are now moving beyond the medium of books
into the medium of electronic hypertext (IV.137). At this stage, our best prospects would be to work from
two ends at once: sizeable teams of researchers designing a data-friendly
lexicogrammar while using detailed analysis and skilfully designed software to
distill regularities from a large computer corpus (II.79; IV.133;
VII.344). This two-way strategy would help us judge the °delicacy of standing
or emergent constraints°, and would keep the description in close contact with
°constraints of language, world, and society°. But it would still remain
incomplete, because the corpus itself is always a partial set of the totality of
English discourse that has been or can be produced. At most, our description
could claim to represent the lexicogrammar of English with a °steadily
finer approximation and with increasing
delicacy° (II.63, 78; III.22, 185; IV.16;
V.2).
30.
These seven
non-deterministic aspects indicate why our categories for labeling the
Word-Classes or Parts of Speech in English discourse should be °heuristic
rather than formally defined° (cf. IV.6, 20ff). Just as ‘squeezing’ the
indeterminacy out of a model makes the relation between the model and its domain
less determinate (cf. Fig. III.12 in III.40), trying to make our set of labels
fully deterministic would oblige our analysis to make arbitrary decisions and to
flatten out substantial delicacy. Instead, a °post-classical description° can
use discourse data for deciding what
degrees of determinism or delicacy are suitable for particular sectors of
the lexicogrammar (cf. II.79; III.173).
31.
Our main cognitive
terms can be derived from our °basic postulates of experience and knowledge°
(III.8, 69f). An Entity is defined as
having an identity and properties, and
being able to persist through an evolution of States and to participate in connectedness.
The basic organization for connectedness is the Process, subsuming a State,
i.e., a stage in the evolution of an Entity, or an Event, i.e., a change of one State to another, plus the Participants,
e.g. Agent and Target, and the accompanying Circumstances,
e.g. Temporality and Locality. In the
State-Process, the main Participant
is the Entity itself, along with its properties and any other Entities to which
it stands in some Relation;
properties believed to be inherent to the Entity are Attributes, while those stemming from an attitudinal judgement about
it are Values. In the Event-Process,
the main Participants are the Initiator
that helps make the Event happen plus several kinds of Targets
toward which it can be directed: a Directly
Affected Entity undergoing the change, an Indirectly Affected Entity undergoing a further effect of the
change, or a Proposition organizing
some content in a °rank-shifted Subordinate Clause° (cf. IV.6, 34). An
Initiator pursuing an Intention is an
Agent performing an Action
of the ‘doing’ type; when an Initiator gets something or somebody else to do
something, ‘doing’ combines with ‘making do’ (IV.82).
A Process wherein Initiator and Affected Entity converge in the same
Participant has the latter as its Medium,
e.g. ‘behaving’ (IV.49).
32.
Our Word-Classes can in
turn be characterized in relation to these cognitive terms for a given language
such as English, starting with the Content Words. An Entity
or Object is typically grammaticalized
as a Noun, and an Event
or Action as a Verb (but cf.
IV.24). Nouns and Verbs are the major Heads
of English Phrases, thus called Noun
Phrase and Verb Phrase, where the
Head is the °functional core and control center° and the Modifiers
are °adjuncts°: for nouns, the adjectives typically
express the Attributes or Values of Entities; and for verbs, the adverbs
typically express the Circumstances of Events and Actions (cf. III.235).
33.
Among the Function
Words, the articles (like
‘the’) typically indicate whether or not the entity expressed by their head
noun is ‘definite’, i.e.,
known, familiar, expected, and so on, while the deictics
(or demonstratives) (like ‘this’) owe their name to their ‘pointing’
quality (IV.143). The prepositions owe their
name to being ‘pre-posed’ at the start of a prepositional phrase,
which normally contains a head noun
or noun phrase and is in turn a Modifier of a noun phrase or a
verb phrase in order to express an Attribute, Circumstance, and so
on. The Junctions (traditionally called ‘conjunctions’, IV.198)
connect nearly all °ranks°, whether they be just two nouns
or verbs, or two entire Sentences
or Paragraphs.
34.
A clause
expresses a Process with a Participant as a noun
or noun phrase
in the Subject standing in a relation
of Agreement (or ‘government’ or
‘concord’) with a Process as a verb or verb phrase
in the Predicate, which may also include one or more Targets as a Direct
Object or Indirect Object. The agreeing
verb is finite
in being formally differentiated by Person, Number, Tense, and, in some
languages, Aspect (e.g. Russian) or Gender (e.g. Arabic). A subordinate
clause is subsidiary
to another Clause and is usually
begun with a subordinating Junction, whereas an independent clause
is not. When a sequence containing at least one independent
clause is marked off as a unit by
intonation or punctuation, the result is customarily called a sentence
— a popular but fuzzy unit that has had a checkered career in the history of
language study (cf. II.39f, 83-86). For many purposes, we can be content with
the Clause and, for a combination of these, the Clause Complex. The Sentence might then be regarded not as the
theoretical unit of language par excellence (as in generative linguistics,
II.83), but as a practical unit of written
language upon which various theoretical units (e.g ‘syntagmemes’) can be
‘mapped’ or projected (cf. II.92).
35. The
main Participants expressed as Subject, Direct Object, and Indirect Object, plus
the Process expressed by the Verb Phrase constitute the Clause
Core. Circumstances such as Temporality
and Locality are usually expressed by Modifiers as Clause Adjuncts. Whether a Process is packaged either in Core or
Adjunct and either in one Clause or several depends on the degrees of focus and
elaboration deemed appropriate
to
the ongoing discourse, e.g., whether the Process gets °topicalized°. The
packaging both reflects and organizes the design of the Process, e.g., how
simple or complex it is. Some expressions are like a momentary snapshot, while
others are like a continuing video. Focus naturally goes more to Event-Processes
than to State-Processes, because changes are naturally more informative and
newsworthy; among the basic changes, temporal
and local are more common and substantial
and dimensional are less so (cf.
III.8).
36.
For Event-Processes, the
most decisive factor in the lexicogrammar of English Clauses is the degree and
mode of °control: what brings about the Process and how?° Yet English is
not fully consistent or clear on this point, apparently shifting gradually
between two perspectives on Clause phrasing. Within the Transitive perspective, the Initiator
and the Affected Entity are clearly
designated, and the Prototype has an °Agent exerting a strong Initiative and
performing an obtrusive hard-coupled Action that causes immediate and
predictable Effects°, as in the linguists’ evergreen ‘the man hit the
ball’ (cf. IV.57, 82, 165).Within the Ergative
perspective, the relationships are more mediated and complex, with the
Initiative coming partly from the inside and partly from the outside or being
distributed among the Participants who make something happen (IV.49f, 82), as in
‘the man hit the roof when we told him the news’. The lexicogrammar of
English Clauses looks unstable because Processes that are cognitively closer to
Ergative are often °grammaticalized° as Transitive, perhaps because the latter
is easier to align with the °classical postulates of knowledge° stated in
III.8 — leaving the Ergativity to be °lexicalized°.
37.
As a result, we may not
find consistent formal distinctions between the degrees of control in Processes,
Initiatives, and Effects. For statements like Halliday’s familiar examples:
[67]
the police exploded the bomb
[68]
the sun ripened the bananas
[69]
the report convinced the committee
the
grammatical similarity on the ‘surface’ obscures the ‘deeper’ diversity
among the lexicalized types of Initiative. The Event involving the ‘police’
made an obtrusive and intentional physical change of substantiality on a chosen
Target, whereas the Event involving the ‘sun’ was unintentional; and for the
‘report’, the intention belonged to the producer(s) trying to effect a
mental change upon the Target. Like all Processes, these presuppose an
appropriate
situationality: somebody (hopefully not the police)
constructed the bomb, the banana tree put forth fruit, the report was compiled
and submitted. Each statement is a selective projection whereby one Process
within a more complex Process is grammaticalized in the prototypical English
Transitive Clause phrasing of Subject-Verb-Direct Object modeled on the
Agent-Action-Affected Entity pattern (cf. IV.39, 48, 50, 83, 93, 195).
38.
In the lexicogrammar of
English, the clearest way to highlight formal distinctions among Process types
by their Initiatives and Effects is to examine the distribution of unmarked
Imperatives. °Commonsense knowledge of world and society° stipulates that
you can normally only command an Action for a genuine intentional Agent capable
of exerting the effort to perform and control it. Hence, [67a] is unmarked,
whereas [68a-69a] would be marked as childish or facetious anthropomorphism.
Negative Imperatives are even more sensitive indicators: such commands imply
that the Action is not just feasible but expected, and that the Agent has the
control to refrain. While the type illustrated by [67b] remains unmarked,
[68b-69b] are even more marked because the prospective Agents are predisposed
neither to act nor to refrain.
[67a]
Inspector, explode the bomb!
[68a]
?sun, ripen these bananas!
[69a]
?okay now, report, convince the
committee!
[67b]
Inspector, don’t explode the bomb!
[68b]
??sun, don’t ripen these bananas!
[69b]
??okay now, report, don’t convince
the committee!
The
English Imperative may have retained its stable constraints because it was not
richly exploited during the historical diversification of social roles and
divisions of labor (cf. III.112, 118;
IV.60). More mediated and adaptable means of command and request were derived
instead from Modal Verb constructions, Interrogatives, and so on (cf. IV.60).
39.
The lexicogrammar of
English is also fairly well differentiated in the distribution of Cleft
constructions, which ‘cleave’ and rearrange the prototypical
Subject-Verb-Direct Object pattern in order to place more focus on some items.
For focusing on Processes or Participants, Clefts with ‘what’ are quite
useful. The Pro-Verb Cleft or Pro-Cleft for
short (e.g., ‘what the man did was hit the ball’) uses the Pro-Form
‘what’ and the Pro-Verb ‘do’ and saves the Process Verb (like ‘hit’)
for a higher-focused slot further on (cf. IV.68, 194, 217). You are told
something is or was ‘done’ by the Agent and wait to find out just what; a
brief uncertainty is created to draw attention and is then resolved. The added
focus again implies effort and control by the Agent, so that [67c] is less
marked than [68c]. But the criteria are less clear-cut than for Imperatives; a
version like [69c] can be used without Initiative in the sense of ‘the overall
effect or result was that…’.
[67c]
what the police did was explode the bomb.
[68c]
?what the sun did was ripen the
bananas.
[69c]
what the report did was convince the committee.
The
constraints are different for the Process-Verb
Cleft or Process-Cleft for short,
which has the Process Verb itself in the place of the Pro-Verb ‘do’ (e.g.,
‘what the man hit was the ball’). The focus goes on the Target, and this
usage can be oddly marked if it is strongly °collocative°
with the Process (e.g., ‘?what the
police exploded was the bomb’), because the uncertainty seems gratuitous (what
else would they explode?) unless a contrast is involved (e.g., ‘what the
police exploded was not the bomb but the myth of their being experts on
terrorism’). In return, Process-Clefts are usually unmarked for non-effortful
or non-controlled Processes (e.g., ‘what she knew was that he was lying’ or
‘what we want is Watney’s’) which strongly resist Pro-Clefts (e.g.,
‘*what she did was know that he was lying’ or ‘*what we do is want
Watney’s’) (cf. IV.81). ‘It’-Clefts, beginning with ‘it’ plus some
form of the Verb ‘be’, are the most versatile of all and serve to focus on
many sorts of Participants and Circumstances (e.g., ‘it was the third man who
hit the ball’, ‘it was the knuckle ball that the man hit’, ‘it was in
the second inning that he got a hit’, etc.) (IV.218).
40.
To further assess the
type of Initiative and the degree of control over it, we can deploy °denial
tests° that reflect cognitive and social constraints upon making denials.
The Denial of Avoidance Test inserts ‘can’t/couldn’t help’
before the Process Verb and is unmarked when the Initiator could reasonably not
have adequate control. It would make sense to say, for instance, ‘the police
couldn’t help exploding the bomb’ (maybe it had a hair trigger) but hardly
‘?the police couldn’t help
shooting the demonstrators’ (they did so at their own discretion and could
have refrained). The Denial of Intention
Test inserts ‘doesn’t/don’t/ didn’t mean to’ and is unmarked for
Processes whose Initiator had some control but didn’t manage to carry out the
Intention. It would make sense to say ‘the police didn’t mean to shoot the
demonstrators’ (their Intention was to fire a warning), but hardly ‘?the
sun didn’t mean to ripen the bananas’ (it didn’t have any Intention either
way). Both tests are sensitive to the discrepancy between what you intend and
what you can actually manage to control, achieve, or prevent, especially where
social conventions are restrictive or choices must be rapidly made among similar
alternatives. But the two tests do not give the same results, as we see from
examples like ‘shooting the demonstrators’. Finally, we can apply a Denial
of Consequence Test that states a Process and then negates another Process
or Proposition that ought to follow as a normal consequence; e.g., ‘??they
exploded the bomb but it didn’t go off’ is quite marked because
‘exploding’ is an Intensive Dispositive Process’ that follows through, as
contrasted with ‘they jiggled the bomb but it didn’t go off’. This test is
familiar for Processes entailing commitment to a Belief: ‘??I
knew they had exploded the bomb but they hadn’t’, as contrasted with ‘I
believed they had exploded the bomb but they hadn’t’ (cf. IV.77, 116).
IV.B.1
Designing a scheme of Processes and Aspects
41. For classifying Processes, we can revise Halliday’s ‘systemic functional scheme’ to attain the °linguistic, cognitive, and social scheme° shown in Table IV.1. Its terms will also be capitalized to
distinguish them from other uses, e.g.,
‘Perception’ and ‘Cognition’ here versus °perception and cognition as
the major human processes whose dialectic° was described in III.85. The scheme
cannot be a final deterministic catalogue whereby every expression can be
exactly pigeonholed under just one Process, because Processes can be richly
shaded and combined by °emergent constraints in ongoing discourse°. The scheme
is merely intended to map out some regular and influential form-function
correlations within English discourse between lexicogrammatical constraints
(including morphological, lexical, and syntactic ones) and cognitive and social
constraints (including semantic and pragmatic ones) from knowledge about world
and society. The ways in which °previously undescribed languages organize
different models of world and society has been strikingly demonstrated by
fieldwork in discourse analysis°, such as the impressive studies carried out by
the Summer Institute of Linguistics (cf. II.37; IV.45).
42.
The first two major
categories in the scheme are Endocentric
Processes featuring ‘inner’ Events (e.g., ‘having a bright idea’),
and Exocentric Processes featuring
‘outer’ Events (e.g., ‘slipping on a banana skin and falling on your
asphalt’). The former type is closer to ‘mental’ or ‘data-based’
activities and the latter type to ‘behavioral’ or ‘material-based’
activities, but we must not imply a dichotomy like the old dualism of ‘mind
versus body’ (III.17). The newer terms indicate where the Process is centered,
i.e., which domain dominates in
activities that are working to coordinate inner with outer, mentation with
behavior, and of course data with material (IV.103f).
The clearest Exocentric Processes are obtrusive,
i.e., can be readily perceived (III.84), and prefer Targets — an influential
and fairly °classical enrichment of the basic postulates of observability,
substantiality, and connectedness° (from III.8).
43.
The third major category
is Representive Processes expressing the °identity and connectedness (e.g.
temporality and locality)° of Entities and States in terms of properties,
relations, or significances rather
than of participation in Events. Thus,
Representive Processes lack the Transitive or Ergative perspective of
Initiatives and Effects, and register or qualify world and society rather than
intervening or targeting (IV.69, 91). They can mediate by Representing what
Endocentric Processes like ‘seeing’ have concluded about Exocentric
Processes like ‘doing’.
44.
The fourth and final
major category is Expressive Processes
that ‘externalize’ inner States, Events, and Representations according to
culturally determined conventions, typically with some Target being the
Addressee of the Feelings, whereas Semiotic Processes express Content and
Messages, though these two Process types often interact richly in discourse (IV.
103, 109, 112, 118).
People who perceive your feelings tend to treat them as Messages; and your
explicit Messages tend to express or affect your feelings or those of your
audience. Still, the lexicogrammar of English reflects the commonsense belief
that feeling and saying are two distinct matters (cf. IV.110).
45.
Aspects, as the term suggests, project perspectives
on how Processes come about: not what happens but how, when, whether, how often,
how long, and so on. Although Aspect signals often appear in Verb Phrases —
for many languages more so than English — the Aspects themselves can apply to
a Process expressed in a whole Clause, Clause Complex, Discoursal Move, or
Discourse Episode. As we know from °fieldwork on remote languages° such as
Wintu (V.44), Aspects can be
systematically grammaticalized to widely varying degrees. English is rather
sparse and unsystematic (IV.47, 55, 57): a few Aspects are grammaticalized by
Auxiliaries and Modals, while most are lexicalized by Adverbials and specific
Process Verbs or by other Word-Classes formed from Process Verbs, such as Nouns
or Participles (IV.149).
So, most traditional grammars of English have eschewed the term ‘Aspect’ and
offered sparing treatments of ‘tenses’, ‘voices’, and ‘moods’ (terms
borrowed from Latin grammar). Still, I submit that the functional Aspects
presented below are normally selected in actual English discourse (even in invented
sentences) but often in °unmarked
ways that are so sparsely grammaticalized° as to be easily overlooked (cf.
IV.57).
46.
The proposed scheme of
Aspects has eight categories (Table IV.2).
Polarity
has the two terms of Positive and Negative.
Logically, they appear evenly balanced and complete, like yes and no, + and -,
or 1 and Ø. Either it happened or it didn’t, says the philosophers’ ‘law
of the excluded middle’. However, the Positive is plainly the unmarked side,
because it can be stated without any particular expectations; and many languages
associate Positives with Ameliorative, and Negatives with Pejorative. The
Negative is more marked in normally implying that the Process should happen but
didn’t (IV.38, 159). Hence, a
common sequence of discourse is a Positive statement followed up by a denying
Negative [70]. The choice among Negative patterns allows a speaker to deny
specific components of a Process, and multiple Negation is common in spoken
discourse, e.g. [71], where the answer denies not merely the Event, but the
Agent and the Circumstance of Locality — a usage condemned by logically
oriented °language guardians° of English on the grounds that Negatives cancel
each other. Yet cancelling is rather marked and uncommon for the
‘“no”-word Negatives’ (‘no, not, nothing, nobody, nowhere’, etc.), e.g., to deny that
you are denying [72] (cf.VII.255). Cancelling is unmarked only if at least one
Negation is some other type of Word (e.g., ‘I don’t deny…’, ‘they
didn’t refuse to…’, ‘it’s not impossible that…’).
[70]
‘She is a very strong-willed selfish person’. ‘No she isn’t.’
[71]
‘Your kid’s come home here’ […] ‘Ain’t nobody come nowhere’
[72] ‘you
wouldn’t recommend me building here waterside?’ ‘Wouldn’t recommend and
wouldn’t not recommend. Neither one. Do what you please.’
In
some languages, e.g. Spanish and Russian, multiple Negation is fully standard,
irrespective of Word types and logical notions.
47.
Tense has traditionally been grasped as a temporal progression going from
Past to Present to Future,
although on close inspection the fit between Tense and Temporality is rather
untidy (IV.155). Tense is concerned
with relative, not absolute
time, and the ‘present’ time when the discourse is taking place ‘now’ is
by no means the sole point of orientation. For example, the Present Tense is
unmarked for Processes that happen a lot, although past and future time are of
course implicated, e.g.:
[73]
The rain in Spain stays mainly on the plain.
If
the terms Past, Present, and Future invoke temporal expanses relative to
‘now’, we can complement them with terms for temporal expanses relative to
each other: Predecessive
(before
that time), Successive
(after
that time), and Progressive
(extending
over time, often in relation to an interceding Process at one point in time). A
Tense with none of these three relational options is conveniently termed Simple.
Compared to many languages, English is fairly sparse in grammaticalizing Tenses
and prefers Auxiliary Verbs over inflections (or ‘conjugations’) of the
Process Verb. Hence, the Simple versions of Present and Past typically contrast
with the non-Simple ones by consisting of just the Process Verb, unless some
other Aspect (e.g. Polarity) needs to be indicated. I shall examine the major
Tenses of English later when surveying the Verb Phrase (IV.147-51);
by then we will have seen that Process type often constrains the distribution of
unmarked Tenses, especially Simple versus Progressive.
48.
The term Transitivity
can cover what has been traditionally, if a bit abstrusely, called ‘voice’
and linked directly to Verb forms, although it concerns the roles of
Participants and Initiatives in an overall Process. The Active
is the least marked category in English, assigning the position of Clause
Subject to the Initiator or Agent as in [67-69] back in IV.37, whereas the Passive assigns
that position to the Affected Entity [67d-69d].
[67d]
the bomb was exploded by the police
[68d]
the bananas were ripened by the sun
[69d]
the committee was convinced by the report
But
(like Positive and Negative) Active and Passive often do not form a balanced or
complete opposition (cf. IV.23). A genuine choice is usually open only for
Clauses expressing a Process with clear Initiative and Effect; and the use of
Active or Passive tilts the focus among the Participants. In Subject position,
the Agent or Affected Entity functions more as the expected Participant (the
‘Thematic’ one, IV.215ff), and in the post-Verbal position more as the
newsworthy Participant (the ‘Rhematic’ one) that merits end weight. Hence,
[75] conveys a belief about the ‘children’ who may learn any three languages, whereas [76a] conveys a statement about the
same ‘twelve languages’. Empirical research indicates that speakers tend
to pick Passives when the Affected Entity outranks the Agent in being larger in
size, more animate, more imageable, perceived earlier, or claiming attention.
Also, speakers can pick the Passive to omit the Agent, e.g., when not known or
relevant in an Expository Perspective [77].
[75]
her parents thought that every child should learn at least three languages
[76]
twelve divergent languages are spoken by the different tribes
[77]
Wood stoves […] required a lot of attention. Dry kindling and green wood had
to be cut to fit the firebox and kept on hand, and the fire had to be watched
[…] (Eliot Wigginton)
When
focus is assigned to some tool or machine, as is done here for explaining how
‘wood stoves’ were tended, the Agents needn’t be specified insofar as they
would all have to perform the same Actions.
49.
The Reflexive combines the roles of Initiator or Agent with Affected
Entity, as in [78]. A Target can be made into an Initiator, e.g., to avoid
naming the real Initiator [79]. The Reciprocal
combines two or more Agents enacting the same Process on each other instead of
on themselves [80].
[78] The Duchess squeezed herself up closer to Alice’s side
[79]
The bacon done burnt itself up
[80]
they fought each other in the gravel by the bus stop
Reflexives
are far less common in English than in many languages, including its near
relatives in the Germanic and Romance families. English prefers the sparser
‘Intransitive’ or ‘non-Ergative’ Medial
in between Active and Passive, where the main Participant in Subject position is
the ‘Medium’ through which the Process takes place (IV.36, 82). The English
Medial serves for many Actions (e.g., ‘get up’, ‘sit down’, ‘lie
down’) which are Reflexives in other languages (e.g., Spanish ‘levantarse’,
French ‘s’assoir’, German ‘sich hinlegen’). When the Reflexive can be
identical in form to the Reciprocal, context can sort them out (in German,
‘sie küßten sich’ is clear enough where kissing oneself is improbable, but
not ‘sie rauften sich die Haare aus’ depending on who tore out whose hair).
50.
The English Medial is
most dominant in the State-Processes featuring the Entity and its Attributes and
Circumstances (e.g., ‘be old’, ‘be here’), and in the Event-Processes
featuring the Entity Enacting or Developing (e.g., ‘move’, ‘become’).
Medial versions seem °unmarked if the Participant is collocative with the Event°,
as in [67e] and [68e], and less so if it is not, as in [69e], though you can
improve the latter through °lexical control° by specifying the conditions and
more strongly implying somebody’s efforts, as in [69f] (cf. IV.23).
[67e]
the bomb exploded
[68e]
the bananas ripened
[69e]
??the committee convinced
[69f]
the committee didn’t convince so easily this time
In
contrast to many languages, the lexicogrammar of English can be adapted to
express a Circumstance as Subject and the Medial Agent as Target (VI.11),
e.g. [81-82] — a trend encouraged by °commodification° and by the
presentation in °consumerist discourse° of inanimate things doing services for
people (IV.238). The prototype of Subject-Verb-Direct Object wins out again
(IV.37):
[81]
the theatre seats 1200 people [vs.: 1200 people are sitting in the theatre/the
ushers seat the audience]
[82]
this trailer sleeps three people [vs.: three people are sleeping in the trailer]
For
some Process types, e.g. Emotives, the Active-Passive distinction is missing and
a distinction falls between an other-directed Ergative Active versus a
self-directed Medial, e.g., ‘enrage’ vs. ‘rage’, or ‘sadden’ vs.
‘feel sad’ (cf. IV.82, 90).
51.
Status
concerns whether a Process is performed, imagined, enjoined, and so on (cf.
IV.58, 160-62). The unmarked Status, the Declarative,
expresses unqualified ‘actuality’, as in the silly aphorism about Spanish
rainfall back in [73]. In the Performative,
the Process converges with the Action of expressing it and inherits its
‘actuality’, e.g. [83]. The Conditional
is used for a contingent Process depending on some condition being met, e.g.
[84]. The Contrafactual is used when
the Process is distinctly hypothetical or fictional, e.g. [85]. Conditional and
Contrafactual enter an unmarked connection when the condition is not met and the
Process can’t occur, e.g. [86]. The Optative
conveys a wish that something would occur though it hasn’t, e.g. [87]. The
older forms to grammaticalize Conditional and Optative, e.g. [88-89], have now
faded to the point where the two rely on the same Modal ‘would’, although
the Optative can also appear without a contingent Process, e.g. [87]. The Imperative
goes furthest by delegating the Initiative to an Agent to make the Process into
an actuality, e.g. [90]. The Interrogative
inquires whether and how the Process occurs or under what Circumstances, e.g.
[91], whereas the Exclamatory
proclaims the occurrence or its Circumstances to be informative, e.g. [92].
Finally, the Subordinate is joined
onto a Independent Clause without sharing its Status as Imperative,
Interrogative, or Exclamatory (IV.26, 28, 159): [90a] isn’t commanding ‘the
boy’ to ‘get here’, [91a] isn’t asking whether ‘you left’, and [92a]
isn’t exclaiming over ‘buying the paintings’. Not even the Declarative
Status of an Independent Clause is shared, because the Subordinate presupposes
or backgrounds something, e.g. that ‘the paintings became very valuable’
[92b] (cf. IV.159, 210).
[83]
This meeting will now come to order!
[84]
if you cut your finger very deeply
with a knife, it usually bleeds
[85]
Had I actually gone through with the ridiculous confession […] I would have
expired of sheer absurdity